ReadonlyvoiceThe voice identifier to use for TTS synthesis. Non-empty string signals that this simulator has a voice channel configured.
Convert text to a voiced ModelMessage using the given voice
config (falls back to the simulator's own defaults when cfg is
omitted).
The cfg parameter accepts a generic structural shape
({ tts?: { voice?: string } }) rather than importing VoiceConfig
from the voice layer — keeping this file in the domain layer without
introducing an upward dependency.
Optionalcfg: { tts?: { voice?: string } }
A voice-capable user simulator that converts text to a voiced ModelMessage (TTS + audio bytes). Used by the interruption path to generate the user barge-in phrase as real audio before feeding it to the agent adapter.